Could a machine become cruel? Perverse? Evil? The answer is yes, and it has been demonstrated by Norman, the world’s first artificial intelligence trained to think like a psychopath.
Its name pays homage to Norman Bates, the protagonist of Psycho, Alfred Hitchcock’s famous horror and suspense film which, since its release in 1960, continues to be acclaimed by international critics as a cinematic work of art. Anthony Perkins had a lot to do with this achievement, as the actor who played the main character, a terrifying serial killer who has gone down in history as the embodiment of a psychopath.
To develop this pathology, they fed “the machine” abundant violent content drawn from such controversial spaces as Reddit, the largest forum on the internet, often criticized for publishing cruel and false information. It is such a controversial platform that it has even apologized for contributing to “making the world a worse place.”
The result is an artificial intelligence that, wherever it looks, sees only death. In fact, it was given the Rorschach test (or inkblot test) to assess its personality, yielding “a chronic hallucinatory disorder” and a strong obsession with murder.
Norman took the test alongside a healthy artificial intelligence and the results left no room for doubt. On one of the cards the “psychopath” saw a person being electrocuted and about to die, while the healthy “mind” perceived a group of birds singing on a branch. On another occasion, Norman recognized a man fatally wounded, while its counterpart saw a vase with a bunch of flowers. During the test, it also made out a pregnant woman falling from a building, a man screaming and dying next to his wife…
Are you already wondering? What is the point of devoting time and money to a project like this? As its creator, MIT researcher Manuel Cebrián, told me the day I met him, it was done “to show that a sick mind can be created in a machine and not only in a human being.” Can you imagine such a “being” conversing on social networks? Controlling a border? Or caring for a human being?
Technology is a tool that has no “natural” goodness or evil, unlike man, but it is advancing at a dizzying pace and, if we do not set limits, as this project has managed to show, the result could be a “demon.” For that reason, every day more governments, scientists and social movements are trying to warn about this danger and, of course, demanding measures. One of the greatest advocates was Stephen Hawking, for whom “humanity is at risk and many of the dangers we have created ourselves.”
Nevertheless, today my reflection is for you: Are you mindful of how you use technology in your daily life? Are you sure you are? Be careful, because the dangers are there, silently waiting for you, and if we keep putting it off, soon it could be too late.
Originally published in Spanish on IANética (2021). English version prepared for TecnoHumanismo.
Related articles
HAL’s Ethics
Ethics. Algorithms are key elements of the services and infrastructures of information societies. Why we must understand how they decide.
Read article →9 min
Ethics of Advanced Robotics and Humanoids
As advanced robotics and humanoids take hold, an essential debate on ethics emerges: how to integrate AI into robots while protecting human dignity.
Read article →6 min
Good Practices in Artificial Intelligence
A standout panel at the 5th Emergentes 2022, organised by AECTA at Las Naves, discussed good practices in AI under the motto “AI and Humanism”.
Read article →3 min